• grue@lemmy.world
    link
    fedilink
    English
    arrow-up
    15
    ·
    9 месяцев назад

    Code’s not enough. The data the algorithm is analyzing (both the training data and live data) has to be public too, in order to actually understand what the algorithm is doing.

    • WamGams@lemmy.ca
      link
      fedilink
      arrow-up
      8
      ·
      9 месяцев назад

      Is there a way to publish the data without harming the users?

      • grue@lemmy.world
        link
        fedilink
        English
        arrow-up
        4
        ·
        9 месяцев назад

        The training data, yes. As for the live data… I would say “yes” in the sense that the service shouldn’t be storing anything but what the users explicitly choose to make public to begin with.

        • rhandyrhoads@lemmy.world
          link
          fedilink
          arrow-up
          6
          ·
          9 месяцев назад

          Do you think there’s a difference between the training data and the live data? Didn’t most of these platforms start without engagement algorithms?